# AgentCanvas

- **Event:** [Google I/O Hackathon](https://cerebralvalley.ai/e/google-io-hackathon)
- **When:** Sat, May 23 at 9:00 AM – 10:00 PM (PDT)
- **Where:** Shack15, San Francisco, CA
- **Team:** [Preeti Parihar](https://cerebralvalley.ai/u/itspparihar)
- **GitHub:** https://github.com/preeti13parihar/agentcanvas
- **Demo video:** https://drive.google.com/file/d/1lVv_6AHcWWv3UKFoBJoAgd-ppVmLxyrr/view?usp=sharing
- **Gallery:** https://cerebralvalley.ai/e/google-io-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/google-io-hackathon/hackathon/gallery/6

AgentCanvas is the first visual orchestration canvas for Gemini 3.5 Flash sub-agent pipelines. Most multi-agent systems today are black boxes - you run code, agents execute, you see a final output with no visibility into what happened in between.
AgentCanvas changes that. Drag and drop Gemini agents onto a canvas, connect them into pipelines, define their roles and tools, and watch every tool call execute in real time with a live animated trace. Every agent handoff, every tool call, every output is visible as it happens.
Built entirely during the hackathon. Three demo pipelines included: Research (Scout → Analyst → Writer), Productivity (Planner → Scheduler → Communicator), and Engineering (Researcher → Engineer → Reviewer). Fully customizable add agents, define roles, draw connections, run any pipeline.
Tech stack: React frontend with live SSE streaming, FastAPI backend, Gemini 3.5 Flash for all agent execution.

---

Markdown version of https://cerebralvalley.ai/e/google-io-hackathon/hackathon/gallery/6. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt